loss_function_gcn: GCN Loss function

Description Usage Arguments Value

View source: R/loss.R

Description

Loss function for optimising GCN using either AE or VAE

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
loss_function_gcn(
  VAE = FALSE,
  logvar = NULL,
  preds,
  labels,
  mu,
  n_nodes,
  norm,
  pos_weight
)

Arguments

VAE

Whether to use variational auto-encoder or not. Default False.

logvar

Used when VAE is set to TRUE. Default is NULL when using AE.

preds

Predictions made by GCN during training

labels

True values to checked against preds to calculate loss

mu

Mean of the Gaussian for auto-encoder

n_nodes

Number of nodes in the graph

norm

Norm

pos_weight

Weight parameter used to calculate loss

Value

Logistic-cross-entropy loss

#' @references \insertRefscGNNscRGNet


ff98li/scRGNet documentation built on Jan. 14, 2022, 4:58 a.m.